rsync moves one stream at a time. zx Single-Site spreads a single job across many parallel workers, so local copy and sync finish in a fraction of the time — about 10× faster than single-stream rsync. Same unified zx data mover, focused inside one site.
zx Single-Site is part of the unified Zettar zx data mover — the same scale-out engine, pointed at copying and synchronizing data within a single site. rsync was built for one host and one stream; that ceiling turns a cluster-scale copy into a multi-hour wait.
zx Single-Site removes the ceiling. It drives many workers in parallel across self-organizing, self-healing peer-to-peer clusters, running at line rate and scaling out as you add hardware — so the datasets your storage and compute now produce move and stay in sync without the wait.
One copy or sync job is split across many concurrent workers on a scale-out, self-organizing peer-to-peer cluster. Throughput grows with the hardware you add instead of bottlenecking on a single stream.
Workers saturate fast local storage and fabric at line rate rather than crawling file-by-file — roughly 10× the throughput of typical movers, so transfers finish in a fraction of the time.
Each pass copies only what changed and closes the gap until source and target match. Keep large datasets synchronized with repeated incremental runs instead of full recopies.
Built for the file counts, capacities, and metadata loads of cluster- and parallel-file-system-scale local data, with self-healing clusters that keep running where conventional tools fall apart.
"Zettar moved an actual petabyte over a 5,000-mile network loop in 29 hours — with encryption and checksumming — at 96% bandwidth utilization."
Scale-out local copy and sync within a single site or datacenter — a far more efficient rsync for cluster-scale data, part of the unified Zettar zx data mover.
rsync was built for one host and one stream; that ceiling turns a cluster-scale copy into a multi-hour wait. zx Single-Site splits one job across many workers on a self-organizing, self-healing peer-to-peer cluster, so throughput grows with the hardware you add.
Your local LAN speed and storage throughput determine the final rates. Scaling out as you add servers with no software ceiling — the same engine behind the 1 PB record.
Yes — it copies and synchronizes incrementally to convergence, so only what changed moves and copies stay consistent.
Every transfer is protected by unconditional end-to-end checksums, so nothing is silently corrupted or lost.
See how zx Single-Site moves and synchronizes cluster-scale data within your site — scale-out, incremental to convergence, and far more efficient than rsync.